LTC1298 to BS2

Fairuz

Hie...

I am currently designing an optical sensor system and using LTC1298 as part of my design. When I measure the input of my ADC (output of optical sensor)·using multimeter, the measurement shows a very constant voltage until unit of milivolt. However, when I connect the output of my ADC to BS2 and debug the result on screen, the measurement shows signal that keep fluctuating about 20 to 30 steps up and down. I wonder why this happen whereas the measurement using multimeter shows other wise. I hope to receive any suggestion on why this may happen. And how to overcome it, to get a smooth signal. Is there any program line·that may help?

Thanks so much
Fairuz

Beau Schwabe

Fairuz,
It's probably noise in your supply voltage.· Do you have a capacitor (10uf or so) across the power supply pins to the LTC1298?

At 12-Bit resolution, that's about 1.22 mV per step.· To see a value fluctuate 20 to 30 steps only represents 24.4mV to 36.6mV of fluctuation. Better filtering on the supply voltage to the LTC1298 is the only way to reduce or eliminate this, but in most applications, anything under 50mV is pretty good.· If the Capacitor does not work, you can further isolate any power line fluctuations by using a diode in combination with a capacitor to act as a filter.· The problem with that though, is now you have a diode drop (typically about 0.6V) as part of your supply to the LTC1298.· So instead of 5V it "sees" 4.4V ...

                         Diode
Main Positive Supply >----->|-----o--------> Supply to LTC1298
                                  |  + 
                                  o---||---> GND
                                   10uF CAP

... The lower Reference voltage could be compensated for in software, but if you didn't want to go that route, another alternative would be to have a voltage regulator specifically dedicated to the LTC1298.
·
Here is one, perhaps overkill for your application, but it works well when the supply line is VERY noisy.
http://forums.parallax.com/attachment.php?attachmentid=39280



ALSO....

The supply line to your optical sensor may also need a "quiet" power supply, otherwise, noise generated there could make its' way into the LTC1298



▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Beau Schwabe

IC Layout Engineer
Parallax, Inc.

Post Edited (Beau Schwabe (Parallax)) : 5/21/2008 4:29:46 AM GMT

Tracy Allen

A couple of other things might be going on as well.

Multimeters are often very good at rejecting fluctuations at the power line frequency, and if your optical system is picking up noise at 60 or 50 Hz, it would be possible to see those fluctuations on the LTC1298 and even though the multimeter is relatively steady. The cure for that is lowering the source impedance of the optical sensor, or using shielding/twisted pair wiring. Averaging in firmware should be the last step.

It is important where the grounds are connected. The ground from the sensor should be connected as close as possible to the LTC1298 ground pin, not somewhere far off on the other side of the circuit board. Otherwise high frequency noise from the µP can be coupled into the measurement, and the LTC1298 as a sampling converter can see that, whereas a multimeter as an integrating converter will not see it.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Tracy Allen
www.emesystems.com

Fairuz

Thanks so much for the suggestions. It is extremely useful to me.

Thanks
Fairuz